About lizzyAI
lizzyAI is an agentic recruiting platform that helps hiring teams keep work moving across screening, coordination, interviews, and review.
Lizzy takes on repetitive workflow work and turns interactions into structured evidence, while people remain responsible for decisions. We are building recruiting workflows that stay clear, reviewable, and human from the candidate conversation to the recruiter’s next step.
About the role
You will shape the AI behavior behind structured candidate conversations and recruiter-ready evidence. The role combines applied language-model work with product judgment: outputs should be useful, traceable, and easy for a recruiting team to review, correct, or stop.
You will work on the full path from an interaction design or evaluation question to production behavior across voice, video, chat, and mobile channels. The challenge is not only to make a model capable; it is to make the surrounding system measurable, resilient, and honest about what it knows.
What you’ll do
You will pair experimentation with the discipline required for production recruiting workflows. That means defining what good looks like, measuring it against realistic scenarios, and improving both model behavior and the product systems around it.
- Develop and evaluate conversational workflows across phone, video, chat, WhatsApp, SMS, and other mobile recruiting channels.
- Turn role requirements and candidate interactions into structured, source-linked evidence for recruiter review.
- Create evaluation datasets, rubrics, regression checks, and observability for prompts, models, tools, and agent behavior.
- Design guardrails and recovery paths for uncertainty, missing context, model failures, and sensitive candidate interactions.
- Improve how Lizzy communicates its state, source evidence, limitations, and the next action available to a person.
- Partner with product and engineering to move prototypes into reliable workflows and learn from their production behavior.
What you’ll bring
We are looking for applied judgment rather than model novelty for its own sake. You should be comfortable testing assumptions, inspecting failures closely, and explaining quality in terms that product and engineering teammates can act on.
- Experience shipping applied machine-learning, language-model, or agentic systems into production.
- Strong Python skills and comfort with model APIs, tool use, evaluations, experimentation, and production data.
- An evidence-led approach to prompt design, quality measurement, failure analysis, and model or provider selection.
- The ability to balance latency, cost, reliability, safety, and user experience when choosing an implementation.
- Care for candidate experience, clear communication, privacy, and human ownership of hiring decisions.
- Curiosity about voice, multimodal interaction, multilingual behavior, and structured recruiting workflows.
